Journal of Cancer Research and Therapeutics Medknow Publications on behalf of the Association of Radiation Oncologists of India (AROI) ISSN: 0973-1482 EISSN: 1998-4138 Vol. 6, Num. 2, 2010, pp. 185-193

Journal of Cancer Research and Therapeutics, Vol. 6, No. 2, April-June, 2010, pp. 185-193

Original Article

Exploring new potentials and generating hypothesis for management of locally advanced head neck cancer: Analysis of pooled data from two phase II trials

Kundan S. Chufal, Madhup Rastogi ¹ , Sudhir Singh² , M. C. Pant ² , Madhu Srivastava² , M. L. B. Bhatt ²

Departments of Oncology, Batra Hospital and Medical, Research Centre, New Delhi,
¹ Radiotherapy, Indira Gandhi Medical College, Shimla, H.P,
² King George’s Medical, University, Lucknow, U.P, India

Correspondence Address: Dr. Kundan Singh Chufal, Department of Oncology, Batra Hospital and Medical Research Centre, 1, Tughlakabad Institutional Area, New Delhi - 110 062, India.

kundan_25@rediffmail.com

Code Number: cr10041

PMID: 20622366

DOI: 10.4103/0973-1482.65239

Abstract

Keywords: Chemoboost, cisplatin, concurrent, hypoxia, locally advanced head and neck cancer, late chemointensification, mitomycin c, nodal density, radiotherapy

Introduction

Concurrent chemoradiotherapy with better radiation delivery techniques have shown tremendous improvement in the management of locally advanced head neck cancer (LAHNC). ^[1],[2] With the advent of targeted therapies like monoclonal antibodies, the scope for improvement is even bigger. ^[3] At present, the standard management for LAHNC is concurrent chemoradiotherapy. The full potential of this modality can be realized by using it judiciously. The choice of drug and schedule of drug delivery with relation to radiation can impact local control as well as the toxicity profile of the respective radiochemotherapy protocol. The concept of using chemotherapy late in the course of conventionally fractionated radiation, also known as late chemo-intensification protocol (LCI), ^[4],[5] seeks inspiration from the tested logic of the concomitant boost radiotherapy technique. ^[6],[7] This concept has been tested earlier and was found to be feasible in terms of toxicity and efficacy. ^[8] Targeting hypoxic cells with Mitomycin-C (MMC) and improving the overall response to radiation, seems to be a logical approach while dealing with LAHNC. Mitomycin based protocols (MMC) have been used by many authors. Excellent tumor control and survival data have been reported. ^{[9],[10],[11]} We have reported the results of MMC and LCI protocol earlier ^[8],[9] and now we intend to present mature data from both these trials. Pooled data from both these trials has also been analyzed for the purpose of generating a hypothesis.

Materials and Methods

Study design

Between April 2001 and May 2003, two Phase II concurrent chemoradiotherapy protocols were started in our department. Before the enrollment of patients, the institutional review board and clinical research committee approved the trial. A written informed consent was obtained from each patient before his/her participation in the study. In MMC protocol, Mitomycin C and 5FU were given concurrently with radiation. Sixty nine patients were evaluable for final analysis. LCI protocol was late chemointensification with 74 patients evaluable for final analysis. Eligibility criterion was similar for both these protocols. All patients with primary tumors of Oropharynx, Larynx and Hypopharynx with Stages III and IV (M0) disease according to AJCC Cancer Staging Manual, Fifth Edition were included in these studies. All patients had clinical examination, high resolution contrast enhanced computed tomography (HRCECT) of head and neck region, flexible endoscopy, X-ray chest PA view and USG abdomen to assess the primary status of the disease and to rule out distant metastases. Patients were required to have pathologically confirmed squamous cell carcinoma, a Karnofsky performance status of 70-100, an adequate enteral diet, adequate bone marrow reserve (leukocyte count> 4 x 10 ⁹ /l and platelet count> 10 x 10 ⁹ /l), a normal aspartate aminotransferase and bilirubin levels, a creatinine level < 1.2 mg/dl. Patients with oral cavity tumors were not included. Patients with multiple primary carcinomas were excluded from this study, as were patients whose tumor represented a second primary carcinoma. All patients were evaluated by a head and neck surgeon, radiation oncologist and an otolaryngologist at King George's Medical University.

In the MMC protocol (for detail refer to earlier publication ^[9] ), all patients were treated with continuous intravenous infusion of 5FU (600 mg/m ² for Days 1-5 (120 h) and MMC (10 mg/m ² intravenously) on Day 5 during the first week of radiotherapy and on Day 36, i.e. on the 26 ^th fraction. Radiotherapy commenced a minimum of three hours following the start of 5FU infusion. All patients received a conventionally fractionated irradiation, i.e. 2 Gy per fraction five times a week to a total of 70 Gy in seven weeks in two phases. In the LCI protocol (for detail refer to earlier publication ^[8] ), all patients were treated with continuous intravenous infusion of 5-FU (400 mg/m ² for Day 1-5 (120 h infusion) Week 6 and 7) and Cisplatin (10 mg/m ² intravenously Day 1-5, Week 6 and 7) during the last two week of radiotherapy. After first 10 patients, dose of 5-FU was reduced to 350 mg/m2 because of acute toxicity. Radiotherapy commenced within three hours following the start of 5-FU infusion. Radiation details were as per MMC protocol.

Radiation treatment

All patients underwent simulation using a Tele simulator (Schimadzu) with the neck in neutral position. Immobilization was done with the help of orfit cast. The patients were planned using two fields; parallel opposed lateral portals, by SSD technique. Treatment was delivered via Telecobalt Unit (Therateron 780C AECL Ottawa) with dose normalized at tumor center. Dose homogeneity requirement was 95-105% of the specified centrally absorbed dose, as mentioned in the ICRU 50 reference point. Tissue equivalent compensators were made for all the patients. Two-dimensional computer planning was done on TSG Radplan software. The first phase delivered a total dose of 46 Gy in 23 fractions by parallel and opposed fields to the primary and whole neck including spinal cord within this volume. The second phase treatment was given by reducing the marks anteriorly to save the spinal cord (off cord planning) and clinically negative neck received a total dose of 54 Gy at 2 Gy per fraction. Primary disease and clinically palpable nodes received 70 Gy dose at 2 Gy per fraction. Residual nodes overlying the spinal cord were treated with oblique field plans after 46 Gy while restricting the spinal cord dose to less than 50 Gy. Patients were reviewed weekly for mucosal and skin reactions. Weekly hemogram was performed during the treatment. All toxicities were scored according to Radiation Therapy Oncology Group (RTOG) criteria. ^[12]

Measurement of nodal density and tumor volume

HRCECT scans of the entire neck region including primary were performed on GE Medical system, machine model CT/e. Iodine-based contrast material (150 ml) was injected intravenously starting with a bolus of 50 ml (3 ml/second) followed by slow (1 ml/second) infusion of the remaining dose. Total tumor volume of the primary and involved neck nodes was calculated as a cuboid volume using maximum dimension in each plane. Nodal density was graded according to the criteria of the institute Gustave Roussy, France. ^[13] The density of node was compared to that of nuchal muscle at the same level. A node was classified as Isodense if one-third or less than one-third of its cross-section consisted of hypo dense zone and as hypo dense if more than one-third of the node's cross-section was found to be hypo dense. The largest visible node of at least one centimeter was chosen for grading purpose.

Study end points and statistics

The purpose of this analysis was to report the long term results of both MMC and LCI protocols in terms LRFS, DFS and OS. Pooled data analysis of these two protocols was carried out to search for variables having a significant impact over survival functions, with special emphasis on nodal density. The data were analyzed using the SPSS version 15 statistical software. Times for endpoints were calculated from the date of registration. Time dependent variables were analyzed using Kaplan Meier methods. The differences in endpoints within the subgroups, identified in the study, were tested using the Log Rank test. Events for Loco regional relapse free survival (LRFS) included first recurrence of disease at local or regional site or persistent disease.

Persistent disease was regarded as a failure on last date of radiotherapy. Events for disease free survival (DFS) included first recurrences at a local, regional or distant site and death because of disease. Events for overall survival (OS) included all deaths. Patients who lost to follow up were counted as events. Pearson correlation coefficient (PCC) was computed to identify any correlation between the variables. For multivariate analysis, Logistic regression and Cox-Regression Hazard model was used to identify the variables having independent and significant impact over survival functions.

Results

Efficacy and tolerance profile of both these trials have already been published. ^[8],[9] In this paper we present our long term results. Patients and tumor characteristics are presented in [Table - 1].

Mitomycin C protocol: Updated results

Median follow-up time for this study was 55 months (SD 18.5). As on January 2008, there were 31 relapses. Loco regional recurrences were 23 and distant metastases were 8. Six patients with loco regional recurrence develop distant metastasis in subsequent follow up. There were two patients who lost to follow-up without disease and seven lost to follow-up with disease. These nine patients were counted as events in survival analysis. At present there are 31 patients (44.9%) who are alive without disease. Kaplan Meier survival analysis revealed five-year LRFS 59.9% and median LRFS time was not reached, five-year DFS 45.5% and median DFS time was 55 months (95%CI 39.9, 70.0) and 5 year OS 49.5% with median OS time of 60 months (95% CI 39.2, 80.8). Significant late toxicities among the surviving patients were Grade III Xerostomia 76%; Grade II or more sub mucous fibrosis was present in 58%. Some form of long term persistent dysphagia was present in 30% of patients. Results of sub group analysis are presented in [Table - 2].

Late chemo-intensification protocol: Updated results

Median follow-up time for this study was 47.5 months (SD 20.9). In January 2008, there were 40 relapses, 30 loco regional recurrences and 10 distant metastases. Five patients with loco regional recurrence develop distant metastasis in subsequent follow up. There were five patients, who were lost to follow-up without disease and 12 lost to follow-up with disease. These 17 patients were counted as events in survival analysis. Presently 30 patients (40.5%) are alive without disease. Kaplan Meier survival analysis revealed five-year LRFS 53.6% and median LRFS time not reached, five-year DFS 41.5% and median DFS time was 45 months (95% CI 32.9, 57.0) and five-year OS 44.4% with median OS time of 45 months (95% CI 31.7, 58.3). Significant late toxicities among the surviving patients were Grade III Xerostomia 69%, Grade II or more sub mucous fibrosis in 43%. Some form of long term persistent dysphagia was present in 41% of patients. Results of sub group analysis are presented in [Table - 2].

Analysis of pooled data

Data from both the trials were pooled for survival analysis and it revealed 5 year LRFS, DFS and OS of 59.4%, 43.5% and 47.1% respectively with a mean follow up time of 43.8 months (SD ± 19.8). Median survival time for LRFS was not reached while it was 52 months (95% CI 38.0, 65.4) and 52 months (95% CI 34.2, 69.7) for DFS and OS respectively [Figure - 1]. Details of subgroup analysis are presented in [Table - 2]. Direct comparison between MMC and LCI protocol revealed no statistical significant difference in terms of LRFS (59.9, 53.6 P = 0.525), DFS (45.4%, 41.5% P = 0.567) and OS (49.5%, 44.4% P = 0.494).

Nodal disease was present in 133 (93%) patients. Distribution as per nodal density revealed Isodense node in 52 (39.1%) patients and hypo dense nodes in 81 (60.9%) patients. Survival analysis among patients with hypo dense nodes revealed LRFS, DFS and OS of 54.2%, 43.9% and 48.8% at five years in the MMC protocol, respectively, while it was 30.8%, 17.7% and 20.2% at five years in the LCI protocol [Table - 2]. For DFS and OS this difference reached statistical significance (P = 0.008 and 0.007 respectively) [Figure - 2] c and d while it was not significant for LRFS (P = 0.09). Patients with Isodense nodes had LRFS, DFS and OS of 67.7%, 47.6% and 50.6% at five years in the MMC protocol while it was 76.7%, 76.4% and 74.9% at 5 year for the LCI protocol [Table - 2]. For DFS and OS this difference reached statistical significance (P = 0.033, 0.032 respectively) [Figure - 2] a and b while it was not significant for LRFS (P = 0.463).

Impact of treatment protocol used was noted in larger tumors i.e. for total tumor volume more than 40 cc. Statistically significant difference favoring MMC protocol in terms of LRFS (16.4% Vs. 63.8% P = 0.033), DFS (4.5% Vs. 45.7% P = 0.000) and OS (4.5 Vs. 49.3 P = 0.000) at five years [Figure - 3] a, b and c was noted. Prolongation of total treatment of more than 51 days had significant impact over LRFS favoring MMC protocol. LRFS at five years among patients with TTT > 51 days was 29.9% in LCI protocol while it was 64.3% for MMC protocol (P = 0.049) [Figure - 3]d. Correlation analysis [Table - 3] revealed that local recurrence was significantly correlated with nodal density (Pearson correlation coefficient (PCC) = 0.332, P = 0.015). Nodal density had significant correlation with TTV (PCC = 0.280, P = 0.001) and Nodal stage (PCC = 0.259, P = 0.002). TTV had significant correlation with Nodal stage (PCC = 0.475, P = 0.000) and T stage (PCC = 0.332, P = 0.000). Among the same nodal stage, differences in terms of LRFS, DFS and OS were present in patients with Isodense versus hypo dense nodes. For N1 stage DFS and OS at five years were 97.1% and 51.4% respectively, in patients with Isodense nodes; it was 22.2% and 27.8% for hypo dense nodes and these differences were statistically significant ( P = 0.028, 0.038 respectively) [Figure - 4] a and b. For N2 stage, LRFS, DFS and OS at five years were 95.7%, 77.2% and 80.7%; for Isodense nodes while it was 43.5%, 29.4% and 32.4% at five years for hypo dense nodes; these differences were statistically significant ( P = 0.000, 0.000, 0.000 respectively) [Figure - 4] c-e.

Multivariate analysis using logistic regression model revealed nodal density as an independent predictor of loco regional recurrence with P value of 0.015. Cox regression proportional hazard model using forward stepwise (likelihood ratio) method for Block 1 including all variables having significant impact on survival functions on univariate analysis using log rank test and Enter method for Block 2 having treatment protocol as only variable was used. It revealed only nodal density as an independent variable having impact over LRFS, DFS and OS (P 5 0.049, 0.000, 0.000 respectively). Cumulative LRFS hazard for hypo dense nodes was 1.6 times of Isodense nodes and cumulative DFS and OS hazard for hypo dense nodes was 2.90 and 2.91 times of Isodense nodes respectively.

Discussion

Efficacy and safety of LCI and MMC protocols have already been published. ^[8],[9] Results of both these protocols were encouraging and are sustained at longer follow-up. Concurrent chemoradiotherapy and innovative approaches have shown encouraging local control rates ranging from 50% to 70% for locally advanced head neck cancer. ^[14],[15] Innovative approaches such as late chemotherapy intensification protocol revealed that 59% of the patients had two-year freedom from relapse reported by Garden et al. ^[5] while Corry et al. ^[4] reported failure free survival of 40% at two years. Our results for LCI protocol revealed excellent LRFS of 53.6% at five years and median LRFS time not reached. Protocol using MMC has reported five-year OS of 49% and five-year LRFS of 57% by Budach et al. ^[10] Haffty et al. ^[11] have shown three-year overall survival of 70%. In our MMC protocol, five-year LRFS and OS were 59.9% and 49.5% and were comparable with above mentioned studies. Considering locally advanced nature of these head neck cancer, our results are promising.

Analysis of variables

Hypo density in CECT and its correlation with hypoxia has been already shown by Eric Lartiaav et al. ^[16] They have measured PO2 values with CT defined track and shown that mean and median PO2 in the area of the tumor considered to be hypo dense are lower than in isodense area. Nordsmark et al. ^[17] have shown clear correlation between hypoxia and clinical response. The relative risk of failure was 3.56 times higher in most hypoxic subgroup. We found the risk of death in patients with hypo dense nodes was 2.91 times that of patients with isodense nodes. Grabenbauer et al. ^[18] studied the impact of nodal density on the treatment outcome. They reported three-year tumor control of 0% and 91% for hypo dense and Isodense nodes respectively for patients receiving concurrent chemoradiotherapy with Cisplatin.

In our study, pooled data analysis revealed LRFS, DFS and OS of 43.4%, 30.3% and 34.1% at five years for hypo dense nodes, while it was 72.8%, 63% and 66.4% at 5 year for Isodense nodes [Table - 2]. These differences were statistically significant (P 5 0.01, 0.000, 0.000 respectively). Subgroup analysis revealed that LCI protocol was more effective for patients with Isodense nodes [Figure - 2] a and b while MMC protocol was more effective for hypo dense nodes [Figure - 2] c and d. Impact of nodal density persisted within the same nodal stage sub grouping. Hypo dense nodes had a poorer outcome compared to Isodense nodes whether they were in N1 or N2 nodal stage subgroup [Figure - 4]. Subgroup analysis was not done for N3 nodal disease since patients with N3 disease and isodense nodes were very few (n 5 4). This intriguing finding clearly suggests that within the same AJCC nodal stage we have two different tumor subpopulations having different biological behavior. The reasoning behind the LCI protocol was to counter the effect of accelerated repopulation. Incorporation of chemotherapy in this protocol was time dependent and thus this regimen was affected adversely by the prolongation of TTT [Table - 2]. This was manifested by the poorer survival of patients treated with this protocol, in whom the TTT exceeded 51 days. A similar adverse impact of prolongation of TTT was not noted in patients treated with MMC [Figure - 3]d. Tumor volume had significant correlation with nodal density [Table - 3] and thus represents the same interaction as seen with nodal density and treatment protocol. Tumor volume of .40 cc had detrimental effect in terms of LRFS, DFS and OS for LCI protocol but not for MMC protocol [Table - 2].

Generating hypothesis

Our findings generate enough oddities to invoke reconsideration in our approach for selecting chemotherapy drugs and schedules for locally advanced head neck cancer patients. This study is a pooled data analysis from two Phase II trials and thus has its limitation. We cannot draw definitive conclusions but it does help us to hypothesize a pragmatic approach to counter the two most basic radiobiological phenomenon namely accelerated tumor repopulation and tumor hypoxia. Both these events are separated in time during the seven-week course of radiotherapy. We have patients with fairly large nodal and primary disease and thus hypoxia is an inherent problem occurring earlier in the course of radiation treatment while accelerated tumor repopulation occurs late, usually after four weeks of start of radiation treatment. A novel way to treat these patients is to develop a protocol addressing both these phenomena simultaneously along the course of radiation. We propose use of two different chemotherapy drugs at different time intervals, along with radiation, to exploit the maximum benefit. MMC protocol in the initial phase of radiation; LCI protocol during later part of the treatment may be utilized as one of the approaches to treat locally advanced head neck cancer [Figure - 5].

Conclusions

1. Measuring nodal density by HRCECT is non invasive and reliable method. Hypo density is associated with poor loco regional control and survival functions.
2. AJCC nodal staging is not a true representative of tumor behavior as within the same nodal stage patients can have different nodal density and thus have variable outcome. Nodal density is an important marker of tumor behavior.
3. On the basis of our findings it could be speculated that for Isodense nodes LCI protocol is more effective while for patients with hypo dense nodes MMC protocol is more effective.
4. LCI protocol is more sensitive to treatment interruptions, nodal hypo density and larger tumor volume as compared to MMC protocol.
5. Multivariate analysis revealed nodal density as an only independent variable having impact over treatment outcome in terms of LRFS, DFS and OS. Future trials addressing locally advanced head neck cancer should use nodal density as one of the stratification factor. Hypo dense nodes had risk of loco regional failure 1.6 times of Isodense nodes while risk of death was 2.91 times of Isodense nodes.
6. Combination of MMC and LCI protocol [Figure - 5] is worth testing in future Phase I and Phase II studies.

References

Copyright 2010 - Journal of Cancer Research and Therapeutics

The following images related to this document are available:

Photo images